11:00 Eastern; 08:00 Project/Pacific Time. bluejeans.com/426716450
Attendees
Notes
Question from John Swinbank — how fast is BF correction? Yusra said 28s on HSC... can we make it faster?
- Merlin Fisher-Levine suggests turning down the number of iterations — “there is a setting you can adjust, don't know how well it'll work”.
- Using better kernels might be faster.
- Both of these seem a lot less likely now that we know maxIter = 10 (rather than 1,000 which is the threshold, I'd mixed them up in my mind).
- Spent lots of time on AuxTel; lots of activity; eventually got it working.
- Now have images that look like they're actually reading out pixels.
- “Delicately” poking at Spectrator around the edges.
- DM-23105 is to get back to the state we were in before merging the latest changes from Jeremy.
- Merlin Fisher-Levine will open a new ticket for getting the Spectractor reduction pipeline working.
- Should be a “week or two” of work.
- If this means get back to where we were, with things nominally working, but having a better idea of what needs doing to make Spectractor do everything it needs to do, then yes, that should be in a week or two. If this was supposed to mean "everything done and dusted" then that's probably a little bit more work, but it's too early to give a decent estimate of that at this point.
- And getting the stack set up on the mountain.
- John Swinbank may be able to help with Unix questions; Merlin Fisher-Levine can follow up with him offline.
- Will be away for a week from 29 Feb.
Andrés Alejandro Plazas Malagón
- Dealing with family health issues; has delayed getting work done.
- Has been working on DM-21221.
- Has been reviewed by Chris and merged.
- Probably with a breaking change; Merlin will ticket.
- And we should worry about the lack of a test suite.
- Now working on restructuring the task to have a proper run method (DM-23159).
- Has been reviewed by Chris and merged.
- Now need to start thinking about persisting the linearizer.
- Needs some input from Chris on DM-23023; will be blocked this week.
- And to get back to DM-21720.
- Replied to comments on DM-19903; waiting for a reply from John.
- Will work on whichever tickets are unblocked at any given time!
- Still planning to clear technical tickets before spending much time on BOT data.
- Thinking about DM-23023; will raise an RFC to agree to the changes here.
- Coming this week.
- Worried about the size of LUTs needed for LSSTCam; would prefer to save polynomial coefficients.
- Chris will ping the Pierres about their thoughts on how to do an empirical linearity correction for LSST.
- Gen3 cp_pipe bias/dark/flats hopefully coming this week (or maybe next); believe that blockers are resolved.
- Reminder to properly use run/runDataRef methods!
- What's the right answer if run needs 1000 visits?
- Ideally, this is never actually necessary...
- Wait for the bias/dark/flat code above to land — shows what a deferred load looks like.